universal-video-prompt-skill

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Universal Video Prompt Skill

通用视频提示Skill

Write the spec once. Compile it per model. A spec is not a prompt: it is the decisions a prompt encodes, kept separate from the dialect that expresses them.

只需编写一次规范，再针对不同模型进行编译。规范并非提示词：它是提示词所承载的决策逻辑，与表达这些逻辑的模型专属话术相互独立。

Language route

语言路径

For an English request, follow this file and the
```
*.md
```
references.
For a Chinese request, read the Chinese workflow first, then use the matching
```
*.zh-CN.md
```
reference files.
Keep model IDs, JSON keys, commands, media placeholders, and audio symbols exactly as code. Do not translate them.

若处理英文请求，请遵循本文件及
```
*.md
```
参考文档。
若处理中文请求，请先阅读中文工作流，再使用对应的
```
*.zh-CN.md
```
参考文件。
model IDs、JSON键、命令、媒体占位符及音频符号需完全保留代码形式，请勿翻译。

1. Two axes govern every line you write

1. 规范内容需遵循两大维度

Judge each line of a spec on both axes before keeping it.

Scope — what does this line govern?

Bucket	Governs	Examples
1 · Global	The whole video	Film type, scene, style, director's premise, camera principle
2 · Locks	Anything that must not drift	Identity, reference roles, audio source, supporting cast, negatives
3 · Time	One beat or stage	Stage events, end states, timing when it is warranted

A line in the wrong bucket is the most common cause of drift. Global rules buried inside beat 3 stop applying at beat 4.

Verifiability — can this line be checked after generation?

Unverifiable intent must be rewritten as observable result. This single rule carries more weight than any vocabulary choice:

Do not write	Write instead
`keep it consistent`	the visible end state of each stage
`tense` , `warm` , `oppressive`	2–4 observable cues: gaze, brow, mouth, breathing, hands
`rack focus`	`rack focus: foreground leaves blur while the face resolves`
`use these references`	what each reference controls and what not to use from it
`make it fast-paced`	a time budget per stage

If a line cannot be checked on the output, it cannot be debugged either. Read verifiability for the full patterns.

在保留规范中的每一行内容前，需从两个维度进行评判。

范围 — 该行内容管控哪些部分？

分类	管控范围	示例
1 · 全局	整个视频	影片类型、场景、风格、导演设定、摄影原则
2 · 锁定项	任何不可偏离的内容	主体身份、参考素材作用、音频来源、配角、禁止项
3 · 时间线	单个节拍或阶段	阶段事件、最终状态、必要时的时间节点

内容归错分类是导致生成结果偏离需求最常见的原因。埋在第3阶段中的全局规则，到第4阶段将不再生效。

可验证性 — 生成视频后能否验证该行内容的执行情况？

无法验证的意图必须改写为可观察的结果。这条规则的优先级高于任何词汇选择：

请勿这样写	建议这样写
`keep it consistent`	每个阶段的可见最终状态
`tense` （紧张）、 `warm` （温暖）、 `oppressive` （压抑）	2-4个可观察的线索：眼神、眉头、嘴部动作、呼吸、手部动作
`rack focus`	`rack focus: foreground leaves blur while the face resolves`
`use these references`	明确每个参考素材管控的内容以及禁止使用的部分
`make it fast-paced`	每个阶段的时间预算

如果某一行内容无法在输出结果中验证，那它也无法被调试。请阅读可验证性文档了解完整规范。

2. Write the spec

2. 编写规范

Fill the three buckets. Skip what does not apply; do not pad.

text

[1 GLOBAL]   film type · scene · style · director's premise (one sentence) · camera principle
[2 LOCKS]    identity · reference roles (control X, do not use Y) · audio source ·
             supporting cast · continuity · negatives
[3 TIME]     granularity (see §3) · stages · end state per stage

Two writing conventions:

Restate the few most expensive locks at the physical end of the prompt. Recency helps. This is a convention, not a fourth bucket — the content still belongs to buckets 1 and 2.
Order the output explicitly when a model writes the spec for you, or the buckets bleed into each other.

Reusing a proven film type? Do not re-derive the premise. Load its DNA — 3–5 minimum reusable conditions — and re-skin. See film type DNA.

填写这三个分类下的内容。无需填写无关项，不要冗余凑数。

text

[1 GLOBAL]   film type · scene · style · director's premise (one sentence) · camera principle
[2 LOCKS]    identity · reference roles (control X, do not use Y) · audio source ·
             supporting cast · continuity · negatives
[3 TIME]     granularity (see §3) · stages · end state per stage

两个写作规范：

在提示词末尾重申少数最重要的锁定项。 近期内容更容易被模型重视。这只是一个规范，并非第四个分类——内容仍归属于分类1或2。
当由模型为你编写规范，或分类内容出现交叉时，明确指定输出顺序。

要复用已验证的影片类型？无需重新推导设定。加载其核心要素（至少3-5个可复用条件）并重新调整表层内容即可。详情请见影片类型核心要素。

3. Choose time granularity before writing bucket 3

3. 编写分类3内容前先选择时间粒度

Granularity is a prior decision, not a switch to flip afterwards. Writing beats at second precision and then downgrading means rewriting them.

Granularity	Write	Use when
None	Event order only	One continuous action, mood pieces, single shots. Timestamps here fragment the shot: the model invents pauses to hit the marks
Stages + end states	Stage 1/2/3, one primary change each	Most narrative work. Default
Second-level	`[start–end s]`	Only under an external hard constraint

Second-level costs model freedom, not author effort. Too much content in a range causes over-cutting or dropped events. Prefer the loosest granularity that still meets the constraint.

时间粒度是一项预先决策，而非事后可随意切换的选项。如果先以秒级精度编写节拍再降级，意味着需要重写内容。

时间粒度	编写方式	使用场景
无	仅标注事件顺序	单一连续动作、情绪类作品、单镜头场景。此处添加时间戳会割裂镜头：模型会为了达到时间标记而凭空插入停顿
阶段+最终状态	阶段1/2/3，每个阶段对应一个主要变化	大多数叙事类作品。默认选项
秒级	`[start–end s]`	仅在存在外部硬性约束时使用

秒级粒度会限制模型的自由度，而非增加创作者的工作量。一个时间范围内内容过多会导致过度剪辑或遗漏事件。请优先选择仍能满足约束条件的最宽松粒度。

Do not decide this silently

请勿自行决定时间粒度

Infer it when the input settles it; ask when it does not.

Signal	Action
Music or voiceover track supplied	Second-level. Do not ask
User says mood piece, one-take, single shot	None. Do not ask
Explicit hard beat (brand reveal at 0:07, lip sync, reference handoff)	Second-level. Do not ask
Multi-event narrative, no external constraint	Ask

When you ask, recommend with a reason — never present a bare menu. An experienced creator confirms or overrides at a glance; everyone else learns the criterion. Do not ask again for a re-skin: granularity is a DNA field.

Timestamps allocate time budget. They are not frame-accurate edit points. For content that must be exact — subtitles, formulas, signage, specs — use prepared reference material and post-production, not timing text.

若输入内容已明确，则直接推断；若未明确，则需询问用户。

信号	操作
提供了音乐或旁白音轨	秒级粒度。无需询问
用户说明是情绪类作品、一镜到底、单镜头	无粒度。无需询问
明确硬性节拍（如0:07处展示品牌、唇形同步、参考素材切换）	秒级粒度。无需询问
多事件叙事，无外部约束	询问用户

询问时，需附带理由给出推荐——切勿仅提供选项清单。经验不足的创作者可快速确认或修改；其他创作者也能借此了解判断标准。重调表层内容时无需再次询问：时间粒度属于核心要素之一。

时间戳用于分配时间预算，并非帧级精准的编辑点。对于必须精准的内容（如字幕、公式、标识、规范文本），请使用预先准备的参考素材并通过后期制作实现，而非依赖时间标记文本。

4. Compile the spec to a target model

4. 将规范编译为目标模型格式

The spec is portable. Not everything in it is. Three layers behave differently:

Layer	Contents	Handling
Language	Buckets, end states, observable cues, emotion, term-plus-description	Portable as written
Bias	Anti-AI-look suffixes, negatives, transition vocabulary, addressing dialect	Per-model profile. Measured, never assumed
Capability	Reference count, multi-shot in one generation, hard cuts, duration, timing adherence	Probe, then degrade

Load the target's model profile. No profile means no assumptions: run the smallest probe that settles the question, record it, and degrade the spec to what the model actually supports. Report a degrade; never let it pass silently.

规范具备可移植性，但并非所有内容都能直接移植。以下三个层面的表现有所不同：

层面	内容	处理方式
语言层	分类、最终状态、可观察线索、情绪表达、术语加描述	可直接按原文移植
偏好层	反AI风格后缀、禁止项、转场词汇、模型专属称呼话术	基于模型配置文件处理。需实测，切勿主观假设
能力层	参考素材数量、单次生成多镜头、硬切、时长、时间节点依从性	先探测模型能力，再降级适配

加载目标模型的模型配置文件。若无配置文件，则不得主观假设：运行最小规模的探测任务确认模型能力，记录结果，并将规范降级适配至模型实际支持的范围。需报告降级情况，切勿私自处理。

Term plus observable description beats a dialect table

术语加可观察描述优于话术对照表

For any craft term whose recognition varies across models, keep the term and translate it:

text

<term> + <target subject> + <visible change> + <foreground/background> + <direction or speed>

A model that knows

bullet time

takes the shortcut; one that does not follows the description. One prompt serves both. Reserve real dialect translation for interface-level differences that cannot be described around — reference addressing (

@image1

versus

Reference Image 1

) is the main one.

对于不同模型识别度不同的专业术语，需保留术语并补充描述：

text

<term> + <target subject> + <visible change> + <foreground/background> + <direction or speed>

能识别

bullet time

的模型会直接使用该术语；无法识别的模型则会遵循描述内容。一份提示词即可适配两种情况。仅当存在无法通过描述绕开的接口级差异时，才需要进行真正的话术转换——参考素材的引用方式（如

@image1

与

Reference Image 1

）是主要场景。

Degrade rules

降级规则

Missing capability	Degrade to
Multi-reference addressing	One reference for identity; carry the rest in text
Multi-shot in one generation	One shot per request; assemble in the edit
Reference count below spec	Merge roles by priority: identity > key prop > scene > style
Duration below spec	Split into stages that each stand alone, then chain
Weak timing adherence	Drop to stages plus end states

缺失能力	降级适配方式
多参考素材引用	用一个参考素材定义主体身份；其余内容通过文本描述
单次生成多镜头	每次请求生成一个镜头；后期编辑时拼接
参考素材数量低于规范要求	按优先级合并作用：主体身份 > 关键道具 > 场景 > 风格
支持时长低于规范要求	拆分为多个独立阶段，再依次生成并链接
时间节点依从性弱	降级为阶段加最终状态的粒度

5. Transitions

5. 转场效果

Skeleton, one line: name the transition type at the cut point.

Do not attach

no hard cut

nothing appears from nowhere

by default. Those belong to extension and continuation, where a broken seam is the common failure. Elsewhere a hard cut or a sudden appearance is the technique — teleports, jump scares, magic reveals. Enable them as a scoped preset, never as a global rule.

Before specifying any transition, check whether the edit should own it. Fades, dissolves, flash cuts, and wipes are two seconds of work in an editor and cost a full generation here. Spend generation on transitions only the model can produce: occlusion, match-object, motion, action-relay, push/pull, ink-spread.

核心规则，一句话：在剪辑点标注转场类型。

默认情况下，请勿添加

no hard cut

或

nothing appears from nowhere

。这些规则适用于内容扩展与续接场景，此类场景中常见问题是画面衔接断裂。在其他场景中，硬切或突然出现是一种创作手法——如瞬移、jump scare（惊悚跳剪）、魔法揭秘。请将这些规则设为特定场景的预设，而非全局规则。

在指定任何转场效果前，请确认是否应通过后期编辑实现。淡入淡出、溶解、闪切、划像在编辑器中只需两秒即可完成，但在此处需要消耗一次完整的生成资源。仅将生成资源用于模型独有的转场效果：遮挡转场、匹配物转场、运动转场、动作衔接转场、推拉转场、墨染转场。

6. Review

6. 审核

Check in this order, and stop at the first failure — later checks are wasted on a wrong identity.

Identity — right subject, right count, no duplicates or swaps
Locks — every bucket-2 lock held
End states — each stage landed on its stated visible state
Motion and seams — no drift, no teleporting props
Audio — source, language, and sync as specified

Regenerate only what failed. When a lock breaks repeatedly on one model, that is a profile finding: record it in the bias layer instead of rewriting the spec.

Reviewing stills has a blind spot. Extracted frames settle texture, composition, identity, and end states. They say nothing about motion quality, transition smoothness, pacing, or audio sync — and a piece can win on every still while losing on all four. Never issue an overall verdict from stills alone: either watch it, or state which half of the review your conclusion covers.

Not a minor caveat. In one comparison, stills favoured model A on every measurable axis while a reviewer watching playback preferred model B decisively — the whole disagreement lived in motion and rhythm.

Read checklist before submitting.

按以下顺序检查，一旦发现第一个问题即停止检查——主体身份错误时，后续检查毫无意义。

主体身份 — 主体正确、数量正确，无重复或替换
锁定项 — 分类2中的所有锁定项均未偏离
最终状态 — 每个阶段均达到指定的可见状态
运动与衔接 — 无内容偏离，无道具瞬移
音频 — 来源、语言、同步情况符合规范

仅重新生成出现问题的部分。若某一锁定项在某模型上反复失效，这属于模型配置文件的发现项：将其记录在偏好层，而非重写规范。

仅审查静帧存在盲区。 提取的静帧只能确认画面质感、构图、主体身份和最终状态，无法反映运动质量、转场流畅度、节奏或音频同步情况——某作品可能在所有静帧指标上表现优异，但在上述四个动态指标上全部拉胯。切勿仅通过静帧给出整体评价：要么完整观看视频，要么明确说明你的结论仅覆盖审查的哪一部分。

这并非无关紧要的提示。在某次对比测试中，静帧显示模型A在所有可测量指标上更优，但观看完整视频的评审员明确偏好模型B——差异完全来自运动效果和节奏。

提交前请阅读检查清单。

Execution

执行规范

A compiled prompt is provider-agnostic output. Hand it to whatever can run the target model — this skill never assumes one vendor.

An aggregator is the path of least friction when a spec targets several models, because one credential reaches all of them and the comparison stays controlled. Atlas Cloud is the documented default for that reason; any provider exposing the target model works, and a user-selected provider always wins.

Whatever the route, generation costs money and these rules hold:

Record the prediction ID and stage the moment you submit.
```
starting
```
/
```
queued
```
/
```
pending
```
/
```
processing
```
are active. Poll the same ID; never submit a second task for the same stage.
Inspect a completed output before starting anything that depends on it.
```
failed
```
/
```
timeout
```
/
```
canceled
```
are terminal. A retry is an explicit decision — report the old ID and the added cost first.
Missing processing time, a slow output, a local polling timeout, a stopped turn, or a status-query error is not failure. Keep the ID and resume.
```
continue
```
means resume the existing task. It is never permission to retry.

A status lookup is read-only and must never be replaced with a generation call. Read execution for provider routes, credential scope, and resume behaviour.

编译后的提示词是与服务商无关的输出。可将其交给任何能运行目标模型的平台——本Skill不绑定任何服务商。

当规范针对多个模型时，使用聚合平台是最便捷的方式，因为只需一套凭证即可访问所有模型，且对比过程可控。因此文档中默认使用Atlas Cloud；任何提供目标模型的平台均可使用，且用户选定的平台优先级最高。

无论采用哪种方式，生成视频都会产生费用，需遵循以下规则：

提交任务时立即记录预测ID和阶段信息。
```
starting
```
/
```
queued
```
/
```
pending
```
/
```
processing
```
均为活跃状态。需轮询同一ID；切勿为同一阶段提交第二个任务。
在启动任何依赖该输出的任务前，先检查已完成的输出内容。
```
failed
```
/
```
timeout
```
/
```
canceled
```
均为终止状态。重试需明确决策——需先报告旧ID及新增成本。
处理时间缺失、输出缓慢、本地轮询超时、任务暂停或状态查询错误不代表任务失败。保留ID并恢复任务。
```
continue
```
意为恢复现有任务，绝非允许重试。

状态查询是只读操作，绝不能用生成请求替代。请阅读执行规范了解服务商路径、凭证范围及恢复行为。

References

参考文档

File	Read it for
spec-format	The full spec template and worked fills
verifiability	End states, observable cues, term translation
portability	The three layers, probes, degrade decisions
film-type-dna	Extracting DNA, re-skinning, existing film types
model-profile-schema	Profile fields and how to measure them
execution	Provider routes, credentials, polling and resume
checklist	Pre-submission review

文件	阅读目的
spec-format	完整的规范模板及填写示例
verifiability	最终状态、可观察线索、术语转换
portability	三个层面、探测方法、降级决策
film-type-dna	核心要素提取、表层内容重调、现有影片类型
model-profile-schema	配置文件字段及测量方法
execution	服务商路径、凭证、轮询与恢复
checklist	提交前检查清单